Skip to content

Gemini 3.8 Flash

GoogleProprietaryReasoningVerifiedReleased Sep 2, 2026Input: text, image, video, file, audio
Compare with GPT-5.5Source scores ↗OpenRouter ↗
Overall score
69.2
Rank 21 of 222 · 63.6–74.8
Context window
1M
1,000,000 tokens
Price per 1M tokens
$0.75 in · $3.75 out
✓ Verified on Google's pricing page
Blended price
$1.50
3:1 input:output tokens
Output speed
—
Not measured yet

Category scores

Scores per capability area with the model's rank among ranked models in that category.

  1. 1Multimodal#987.3
  2. 2Knowledge#1380.6
  3. 3Coding#1977.5
  4. 4Reasoning#1474.3
  5. 5Agentic & tool use#1672.6
  6. 6Math69.7
Higher is better. #N is the model's rank in that category, where it is eligible to be ranked.

Benchmark results (28)

As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.

Agentic & tool use

Terminal-Bench 2.189.4#5 of 68
MazeBench4#12 of 42
OSWorld 2.059.0#8 of 24
Terminal-Bench 4.019.10#18 of 18
ApprenticeBench24#6 of 18
Hermes Index38.1#9 of 16
Hermes Bench62.2#14 of 16

Multimodal

Design Arena Website1300#14 of 91
ParseBench72.08#6 of 37
ExtractBench (LlamaIndex)(lower is better)80.71#7 of 24
CharXiv w/o tools86.2#3 of 14
LVBench87.1#2 of 6

Reasoning

ARC-AGI-289.2#9 of 41
ARC-AGI-198.50#1 of 36
ARC-AGI-310.4#5 of 22
ARC-AGI-3 Provider Adapter35.00#4 of 8

Knowledge

LABBench286.2#2 of 7
HLE-Verified54.9#1 of 6
BioMysteryBench (human-difficult)56.5#1 of 6
BioMysteryBench (human-solvable)88.8#4 of 5

Coding

DeepSWE73.8#10 of 50
Bug Hunt Bench18.0#21 of 23
FrontierSWE v219.6#16 of 20

Math

FrontierMath v2 (Tiers 1-3)68.421#25 of 64
FrontierMath v2 (Tier 4)21.951#36 of 51

Instruction following

Gray Swan IPI (15 attempts)(lower is better)5.5#5 of 12

Pricing by source

USD per 1M tokens.

SourceInputOutputCached in
✓ Official (Google)
checked Oct 17, 2026 · gemini-3.8-flash
$0.75$3.75—
BenchLM list price$0.75$3.75$0.075
OpenRouter google/gemini-3.8-flash$0.82$4.13$0.083

Paid tier, prompts up to 200K tokens. Several prices change on Dec 1, 2026 / Jan 1, 2027.

OpenRouter lists input at $0.82; OpenRouter lists output at $4.13, which differs from the official price.

Google lists introductory Gemini 3.8 Flash rates through December 31, 2026: $0.75 input, $0.075 cached input, and $3.75 output per million tokens. Flex and Batch input and output rates are half those amounts. On January 1, 2027, standard rates rise to $1.50 input, $0.15 cached input, and $7.50 output; Flex and Batch rates rise to $0.75 input, $0.075 cached input, and $3.75 output. Cache storage is $0.50 per million tokens per hour through 2026 and $1.00 afterward.

Arena ratings

Human-preference ratings from LMArena. Best-rated variant shown.

Text(High)1499#8 · 37.5K votes
WebDev (coding)(High)1585#35 · 10.9K votes
Vision(High)1315#7 · 4.6K votes

Speed

Measured by us through OpenRouter (median of trials).

Not measured yet.

Related models

Same family

Similar score